htozaki
commented
5 months ago Thank you for your interest for DoqueDB.
Unlike Elasticsearch or OpenSearch, DoqueDB is written in C++, so the management of memory is different from Java. So I cannot answer whether it is memory-dependent or not in your terms.
Though I'm not familiar with Clickhouse or OpenObserve, at quick looking their document, it looks they do special management in searching, explicitly use some kind of data to file so as to handle some kind of data efficiently.
What I can say it now is that DoqueDB has no feature that do something for special database like log files etc. Like other SQL-based databases, all data and indices are stored in memory or page cache in OS dependent way, that would be typically based on LRU algorithm on linux.
I apologize in advance if this isn't an appropriate content for a GitHub issue.
I'm interested in DoqueDB's performance. To understand this better, I'd first like to grasp the key characteristics of the software's architecture.
Specifically, I'm curious whether DoqueDB's structure is:
Fully memory-dependent, similar to Elasticsearch or OpenSearch, or Capable of balancing between memory and disk usage, like Clickhouse or OpenObserve. Any information you could provide on this would be greatly appreciated.